MBone: Internet Atlas Gallery

Mbone Tunnels of the Major Backbone Networks

Tamara MunznerStanford University
Eric Hoffman Ipsilon Networks
kc claffy CAIDA at SDSC
Bill Fenner Xerox PARC

URL: http://www-graphics.stanford.edu/papers/mbone/

Visualization Thumbnail	Analysis
	Data . . . Sources . . . Specification . . . Aggregation . . . Filters Visualization . . . Techniques . . . Key Mappings . . . Tools Evaluation . . . Strengths . . . Issues Credits/Contact Info . . . Authors . . . Contact Info
Full-Size Visualization	Back to Gallery

Additional Content

Internet Atlas Gallery

Mbone Tunnels of the Major Backbone Networks

Tamara MunznerStanford University
Eric Hoffman Ipsilon Networks
kc claffy CAIDA at SDSC
Bill Fenner Xerox PARC

URL: http://www-graphics.stanford.edu/papers/mbone/

Visualization Thumbnail	Credits / Contact Info
	Authors: Visualization Tools by Tamara Munzner and Eric Hoffman Mbone Router Tools and Data by Bill Fenner Data Analysis by Tamara Munzner and kc claffy Contact Info: For further information, send email to: munzner @ cs.stanford.edu
Full-Size Visualization	Back

Internet Atlas Gallery

Mbone Tunnels of the Major Backbone Networks

Tamara MunznerStanford University
Eric Hoffman Ipsilon Networks
kc claffy CAIDA at SDSC
Bill Fenner Xerox PARC

URL: http://www-graphics.stanford.edu/papers/mbone/

Visualization Thumbnail	About the Data
	Data Sources: (Source1): Mbone router data (collected from IGMP messages by mrinfo tool by Bill Fenner) for June 1996. Consists of some of MCI and all of Dartnet. (Source2): Cambridge map file from June 1996. Data generated by mwatch mbone monitoring program (Piet Brooks, University of Cambridge), but monitoring has since ceased due to a change in UKERNA's billing policy. (Source3): generated database name2place (Source4): generated database place2latlong Data Specification: (Data1) NDP format file containing one line for each edge or node. (Data2) High resolution globe model in VRML file format (.wrl). (from Stuart Levy of the Geometry Center using the CIA World Map database) Data Aggregation: (Aggregate1): Hostnames from IGMP messages are mapped to latitude and longitude. Data Filters: (Filter1): May remove nonbackbone tunnels (originally drawn as thin black lines). (Filter2): May remove any of three groups of tunnels where 1) both endpoints are in same major ISP, 2) one endpoint is in major ISP, or 3) neither endpoint is in a major ISP.
Full-Size Visualization	Back

Internet Atlas Gallery

Mbone Tunnels of the Major Backbone Networks

Tamara MunznerStanford University
Eric Hoffman Ipsilon Networks
kc claffy CAIDA at SDSC
Bill Fenner Xerox PARC

URL: http://www-graphics.stanford.edu/papers/mbone/

Visualization Thumbnail	Evaluation
Alternate Visualization Technique (Click image for full-size version.)	Strengths: (Strength1): Data format supports animations as well as interaction via a VRML browser. Clicking on a tunnel in the VRML 3D scene triggers the display of hypertext information about its endpoints. (Strength2): Useful for the detection of anomalies and problems in the distribution framework. Issues: (Issue1): Long arcs often hide interesting details. (Issue2): Lines representing tunnels are too dense on this two dimensional projected map to allow further interpretation. An alternative visualization using a globe-based three dimensional representation to give the display more locality is shown below it.
Full-Size Visualization	Back

Internet Atlas Gallery

Mbone Tunnels of the Major Backbone Networks

Tamara MunznerStanford University
Eric Hoffman Ipsilon Networks
kc claffy CAIDA at SDSC
Bill Fenner Xerox PARC

URL: http://www-graphics.stanford.edu/papers/mbone/

Visualization Thumbnail	About the Visualization
	Visualization Techniques: (Technique1): Convert machine-readable mrinfo data into NDP format. (Technique2): Add latitude and longitude to hostnames in the NDP file using hostresolve.perl (based on hostnames) and trim_bidirect (based on IP address) tools. (Technique3): Computer the shortest geodesic arc between lat/longs of the endpoints of each tunnel. The height of the arc depends on its length. Maintain a minimum arc height so that even short tunnels remain visible. (Technique3): Distinguish three groups of hosts where 1) both endpoints of tunnel belong to ISP, 2) one endpoint is within an ISP, and 3) neither endpoint lies within an ISP. (Technique3): Specify finely subdivided arcs (arcstep=20) on a high-resolution globe (~2MB). (Technique4): Support VRML interaction by generating an OOGL file using edge2gv, then using oogl2vrml converter. Key Visualization Mappings: (Mapping1): Color distinguishes ISPs. (MCI=red, Sprintlink=cyan, NASA=blue, ESnet=green, Dartnet=magenta, BBNPlanet=yellow, ANS=white, all other ISPs= black. (Mapping2): Major providers are shown by thick lines, all other providers are drawn as thin black lines. (Mapping3): Position of hosts is based on latitude and longitude information in the NDP file. Visualization Tools: (Tool1): tunnel2edge.perl (converts mrinfo to NDP format) (Tool2): hostresolve.perl, trim_bidirect (add lat/long) (Tool3): edge2gv (calculate arcs and gen .oogl file, then convert to .wrl) (Tool4): grouper (assigns geom tags such as color based on edge file tags) and linegrouper (same but works across multiple records.) (Tool5): WebOOGL/Geomview system (highlight coloring, linewidth selection, manual tweaking, and animations of mbone tunnel growth)
Full-Size Visualization	Back